How does usage work?

Solo, Pro and Enterprise include a monthly allotment of usage. It is the only meter. AI, enrichment, call recording and spoken practice all draw on it, and there is no separate dollar budget to manage.

What a unit is. One usage unit is one cent of what the work costs us. That makes your allotment a spend ceiling you can read without arithmetic: a 10,000-unit pool is $100 of provider cost at full burn. Light actions are cheap and heavy ones are not: parsing a page is 2 units, drafting outreach is 6, a Legion run 10, a Brain query 11, an account brief 15. An action costs the same wherever you run it from — the app, the extension and the API all charge one price — and bulk work shows the total before it starts.

Enrichment is charged at cost, per lookup, and only for the vendor that actually answered. A data pull is a paid vendor call rather than a model call, so it draws more than an agent does: a full Hog lookup is 144 units because that is what The Hog charges us for it.

Where to see your balance. Settings → Usage. It shows what you have used this month, what is left, and the last 30 days broken down by which feature spent it.

The monthly allotment.

• Basic — AI agents are not included. The CRM, the game layer and the rest of the Basic tier work as they always have; the AI features show an upgrade prompt.
• Solo — 5,000 units a month ($50 of provider spend), included with the seat.
• Pro — 10,000 units a month ($100), sized for running the full agent stack daily alongside recording and practice.
• Enterprise — 30,000 units per seat, pooled across the whole org, plus Brain and the Customer Success agent.

Usage resets on the 1st of every month.

What recording and practice cost. A recorded minute is 1 unit per audio track, so an ordinary two-sided call is 2 units a minute: a half-hour call is 60 units. A spoken practice session is priced by the length you pick, 30 units for five minutes, 80 for ten, 151 for fifteen — it climbs faster than the minutes do, because a longer conversation costs more per minute to hold in the model's head. Both draw on the same balance as everything else, so you can see what is left in one place, in Settings → Usage. A separate daily cap on recording (20 recordings or 5 hours in any 24 hours) still applies as an abuse brake.

Running out. AI pauses when the pool is dry, and nothing is billed automatically. Overage is off unless you switch it on.

If you would rather keep working than wait for the reset, turn on overage billing in Settings → Usage. Work past the allotment is then charged at $0.03 a unit, and only the portion that crosses the line is billed, not the whole action that crossed it. Switch it back off at any time. With overage off, a dry pool pauses recording and practice the same way it pauses an agent.

Failed runs are not charged. If a model is unreachable, returns an error, or hits its own iteration limit without producing an answer, the usage for that run is returned automatically. You do not need to ask.

BYOK (bring your own key). Not available yet. Every model call runs on DealArena's provider keys today, on every plan, and draws down usage. Bringing your own model key is on the Enterprise roadmap.

A note on limits you may have seen before. DealArena used to show a monthly dollar spend cap in Settings, defaulting to $550. That is gone. It measured actions rather than money — every action counted a full dollar regardless of what it actually cost — so it stopped people long before their allotment ran out. This meter replaced it. A separate internal safety limit still exists to stop a runaway automation, but it is set far above normal use and you will not encounter it.

Usage also replaced an earlier "credits" meter, which was an arbitrary $0.04 token. The number now means something: it is the cost.